gov.sandia.cognition.graph.inference

Class SumProductDirectedPropagation<LabelType>

java.lang.Object

gov.sandia.cognition.graph.inference.SumProductInferencingAlgorithm<LabelType>

gov.sandia.cognition.graph.inference.SumProductDirectedPropagation<LabelType>

Type Parameters:

LabelType -

All Implemented Interfaces:

EnergyFunctionSolver<LabelType>

@PublicationReference(author="Tu-Thach Quach and Jeremy D. Wendt",
                      title="A diffusion model for maximizing influence spread in large networks",
                      type=Conference,
                      publication="Proceedings of the International Conference on Social Informatics",
                      year=2016)
public class SumProductDirectedPropagation<LabelType>
extends SumProductInferencingAlgorithm<LabelType>

This class implements a Bayes Net -- but only allowing for pairwise influence along edges. That is p(x | a, b, c) is approximated by p(x | a) * p(x | b) * p(x | c). We realize that's a considerable approximation, but on real graphs with very high degree nodes, it quickly becomes infeasible to compute p(x | a, b, ...) when there are hundreds of incoming edges and some approximation must be chosen.

Author:

jdwendt

Field Summary

Fields inherited from class gov.sandia.cognition.graph.inference.SumProductInferencingAlgorithm

DEFAULT_EPS, DEFAULT_MAX_ITERATIONS, DEFAULT_NUM_THREADS, fn, nodes

Constructor Summary

Constructors

Constructor and Description
SumProductDirectedPropagation() Creates a new pairwise Bayes Net solver with the default settings.
SumProductDirectedPropagation(int maxNumIterations) Creates a new pairwise Bayes Net solver with the default settings excepting max number of iterations.
SumProductDirectedPropagation(int maxNumIterations, double eps, int numThreads) Creates a new pairwise Bayes Net solver with the specified settings.

Method Summary

Modifier and Type	Method and Description
protected void	computeTemporaryMessage(int edge) Private helper that computes the temporary message for the specified edge for the current iteration going in the specified direction

Methods inherited from class gov.sandia.cognition.graph.inference.SumProductInferencingAlgorithm

getBelief, init, solve

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

SumProductDirectedPropagation

public SumProductDirectedPropagation(int maxNumIterations,
                                     double eps,
                                     int numThreads)

Creates a new pairwise Bayes Net solver with the specified settings.

Parameters:

maxNumIterations - The maximum number of iterations that will be run

eps - The stopping epsilon that will be used

numThreads - The number of threads that will be used

SumProductDirectedPropagation

public SumProductDirectedPropagation(int maxNumIterations)

Creates a new pairwise Bayes Net solver with the default settings excepting max number of iterations.

Parameters:

maxNumIterations - The maximum number of iterations that will be run

SumProductDirectedPropagation

public SumProductDirectedPropagation()

Creates a new pairwise Bayes Net solver with the default settings.

Method Detail

computeTemporaryMessage

protected void computeTemporaryMessage(int edge)

Private helper that computes the temporary message for the specified edge for the current iteration going in the specified direction

Specified by:

computeTemporaryMessage in class SumProductInferencingAlgorithm<LabelType>

Parameters:

edge - The edge to compute the temporary message for (you can't replace the current message yet as the current message may be needed for other edges).